AF Ablation Works…But: SHAM PVI

It’s finally here. Twenty years and millions of procedures later, Rajdip Dulai, MBBS, and colleagues in the United Kingdom have presented positive results of SHAM-PVI, the first sham-controlled trial of atrial fibrillation (AF) ablation. We electrophysiologists breathed a collective sigh of relief. 
Many of my colleagues have long denied the need for a placebo-controlled trial. They argue that trials of ablation vs drugs show that ablation is superior in reducing AF episodes. Proponents also argue that, at least in some patients, pulmonary vein isolation (PVI) terminates AF while the fibrillation persists in the muscle bundles within the vein, thereby “proving” that PVI results in immediate cure. 
I offer three rebuttals:
What about the patient who was highly symptomatic before the ablation procedure and now feels great — but is in persistent AF ? This patient clearly has had a placebo effect.
Why is there such poor temporal correlation between AF and stroke? Further, how much AF, in terms of duration and frequency, is disease modifying? The ARTESIA and NOAH-AFNET 6 trials couldn’t tell us the duration of AF that warrants oral anticoagulation. Perhaps AF burden is just something we can measure, a marker of abnormal atrial substrate, and we are ablating a symptom rather than treating a disease..
Finally, if the main reason to ablate AF is to improve the subjective endpoint of quality of life (QOL), you need a placebo control arm. History is replete with examples of placebo controls overturning medical norms.
I have never believed that AF ablation was a complete placebo. My question was how much of it was resistant to placebo effect. 
Cardiac resynchronization therapy (CRT) for patients with heart failure and left bundle branch block improves QOL, but when systematically studied in cross-over trials, the placebo-resistant effect is only 16%. That CRT symptom relief is more than 80% placebo effect should have made us humbler about AF ablation. 
In two UK centers, 126 patients with symptomatic AF were randomly assigned to cryoballoon PVI or a sham procedure wherein patients had femoral venous access and phrenic nerve pacing (as is done in cryoballoon ablation). Cardioversion was done in both groups for the 80% of patients who had persistent AF. 
Patients were 67 years old on average, mostly male (71%), and had AF for more than 2 years. All patients had loop recorders placed before randomization so that AF burden could be assessed. The primary endpoint was AF burden at 6 months. Secondary outcomes included QOL measures, time to events, and safety. 
The absolute change in AF burden from baseline was 60% in the ablation arm vs 35% in the placebo arm, which was a highly significant difference. 
The authors measured QOL with three different scores. All three improved significantly more in the ablation arm. 
The blinding index indicated that patients did not know their treatment assignment after the procedure, but nearly half the patients in the ablation arm had (correctly) guessed their treatment assignment after 6 months. 
There were no serious complications in either arm. 
In the trial’s publication in the Journal of the American Medical Association, the authors concluded that PVI resulted in a reduction in AF burden and improvement in symptoms and QOL compared with a sham procedure. 
That conclusion is fair, but during the presentation at the European Society of Cardiology meeting, Dulai added that the study demonstrated “no clinically relevant placebo effect with PVI.” 
I take issue with that statement. First, you need a third arm, wherein patients received neither PVI nor sham PVI, to determine the true placebo effect. Patients in the sham arm had a 35% reduction in AF burden, and (by visual inspection of the Kaplan-Meier curves) approximately 1 in 5 patients in the sham arm had no AF in follow-up. This may be due to the effect of cardioversion, regression to the mean, or placebo effect. 
Patient selection. SHAM-PVI investigators chose patients most likely to benefit from PVI. These were highly symptomatic patients (80% persistent AF) who had AF for 2 years before study entry. Two years is enough time to become acquainted with symptoms, and it is also enough time to exclude patients whose AF resolved on its own. This matters because there is a now a movement to ablate AF earlier. Given the high rates of AF regression in short-duration AF, a control arm in a placebo-controlled trial that recruited patients with newer-onset or less burdensome AF may perform even more impressively than the current 35% reduction in AF burden seen in SHAM-PVI. 
Choice of sham. The investigators chose a very safe sham control. Sedation, venous access, and phrenic pacing minimizes risk to the placebo group. The limitation is that it did not involve left atrial access via a transseptal puncture. Transseptal puncture induces a small iatrogenic atrial septal defect immediately post-procedure. This often resolves over weeks, but the left-to-right shunting could modify left atrial pressure and improve dyspnea. Such an effect may seem unlikely, but we don’t really know how PVI works; more than half of patients without AF after a procedure have reconnected veins. Thermal ablation may also induce neuromodulation that could affect symptoms after the procedure. The placebo procedure could have included ablation in the right atrial ganglionated plexi. 
Choice of endpoint. Much of the online criticism of this trial centered on the choice of AF burden as the primary endpoint. I agree that this choice guaranteed a positive trial, as numerous studies have shown that PVI reduces AF burden over meds. That is not in question. The question was whether QOL improves over a proper placebo control. And it did. I would propose that it doesn’t matter that QOL was a secondary endpoint. 
Modest differences and patient perceptions. Patients in the sham arm of this trial improved. They had a 35% reduction in AF burden, and 60% of patients had no symptoms (EHRA symptom class I) or mild symptoms with no effect on daily activities (class II) at 6 months. 
I had AF that resolved without a procedure over a year. At that time, my main worry was not AF but a deadly complication of ablation, such as atrial-esophageal fistula. Though these are rare events, the memory of seeing even one young person die from ablation sticks in your brain. After SHAM-PVI, an EP colleague wrote this to me via email: 
Of course I’m not saying sham is the way to go. But if I had a choice of zero risk + 35% decrease in AF vs some risk of fatal event + 60% AF decrease then I might choose the first option for me or my relative. 
The SHAM-PVI investigators and patients deserve our respect. Doing this experiment more than 20 years into our AF ablation experience, and in the face of great enthusiasm to increase the number of these lucrative procedures, was bold and brave. They have shown that, indeed, there is a placebo-resistant effect of PVI. 
But the nuances of this trial leave important questions. There is surely a placebo effect from AF ablation. The question is, how much of the QOL benefit is due to AF reduction and how much is due to the caring signal of having a big procedure? 
I would have never thought that the benefit of CRT comes mostly from placebo effect. 
The good news is that more placebo-controlled trials in AF ablation are coming. Taken together, we will better understand why patients feel better after the procedure. 
But — we should have been curious enough to do these studies before millions of patients had these invasive procedures. 
John Mandrola practices cardiac electrophysiology in Louisville, Kentucky, and is a writer and podcaster for Medscape. He espouses a conservative approach to medical practice. He participates in clinical research and writes often about the state of medical evidence.